<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[fangting的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/yagreenhand</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; yagreenhand]]></copyright><item><title><![CDATA[Python 直接赋值、浅拷贝和深度拷贝解析]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/107038023</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/107038023</guid><author>yagreenhand</author><pubDate>Tue, 30 Jun 2020 11:16:15 +0800</pubDate><description><![CDATA[1、b = a: 赋值引用，a 和 b 都指向同一个对象。

2、b = a.copy(): 浅拷贝, a 和 b 是一个独立的对象，但他们的子对象还是指向统一对象（是引用）。

b = copy.deepcopy(a): 深度拷贝, a 和 b 完全拷贝了父对象及其子对象，两者是完全独立的。

#!/usr/bin/python
# -*-coding:utf-8 -*-
 
import copy
a = [1, 2, 3, 4, ['a', 'b']] #原始对象
 
b = a           ]]></description><category></category></item><item><title><![CDATA[scrapy爬虫学习]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/106597773</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/106597773</guid><author>yagreenhand</author><pubDate>Wed, 10 Jun 2020 18:45:54 +0800</pubDate><description><![CDATA[因毕设需要做数据集，学习爬虫。
1.目前遇到问题，中文description无法返回。
xpath用法，
https://blog.youkuaiyun.com/qq_33472765/article/details/80672281
xpath是从网页返回内容里边去提取的，网页返回的content和浏览器的element可能会有差异，当xpath无法正常提取内容时，我们首先可以将网页请求返回内容保存到本地在顺着xpath路径查看原因。
print(response.body)


...]]></description><category></category></item><item><title><![CDATA[fewshot代码实现]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/106427781</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/106427781</guid><author>yagreenhand</author><pubDate>Fri, 29 May 2020 16:17:54 +0800</pubDate><description><![CDATA[#参考阿里fewshot论文，用于金融场景对话意图判别
https://github.com/zixufang/few_shot

]]></description><category></category></item><item><title><![CDATA[pytorch tensor（.data,.clone(),.detach()）区别]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/104748886</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/104748886</guid><author>yagreenhand</author><pubDate>Mon, 09 Mar 2020 12:39:20 +0800</pubDate><description><![CDATA[clone()是不内存共享的。
clone()得到的Tensor不仅拷贝了原始的value，而且会计算梯度传播信息，copy_()只拷贝数值。对b处理，grad会返回到a身上。但是b 不会有 grad 属性。
b = a.clone()#a经过克隆得到b，
c = (b ** 2).mean()#
c.backward()
print('a_grad:', a.grad)#但是梯度传播没有因此中断...]]></description><category></category></item><item><title><![CDATA[pytorch损失函数总结]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/104622034</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/104622034</guid><author>yagreenhand</author><pubDate>Mon, 02 Mar 2020 22:38:03 +0800</pubDate><description><![CDATA[感谢博主https://blog.fanyangmeng.com/deep-learning/pytorch/pytorchnotes-lossfunc/

]]></description><category></category></item><item><title><![CDATA[pytorch导入预训练模型部分参数]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/104589112</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/104589112</guid><author>yagreenhand</author><pubDate>Sun, 01 Mar 2020 11:39:40 +0800</pubDate><description><![CDATA[checkpoint = torch.load(checkpoint)
        start_epoch = checkpoint["epoch"] + 1
        #best_score = checkpoint["best_score"]
        best_score = 0

        print("\t* Training will conti...]]></description><category></category></item><item><title><![CDATA[multi-teacher学习]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/104394008</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/104394008</guid><author>yagreenhand</author><pubDate>Wed, 19 Feb 2020 21:51:29 +0800</pubDate><description><![CDATA[多老师学习：
1.Learning from Multiple Teacher Networks
1)不同example 输入到网络会有不同的输出，（x1,x2,x3）得到（p1,p2,p3）,(q1,q2,q3)，怎么保证||q1-q2||和||q1-q3||的相对距离（q2和q3谁取胜）和||p1-p2||和||p1-p3||的相对距离相近（p2和p3谁取胜）。
2)如果是多个老师的话，因为不...]]></description><category></category></item><item><title><![CDATA[documentQA文章]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/103795402</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/103795402</guid><author>yagreenhand</author><pubDate>Wed, 01 Jan 2020 19:50:52 +0800</pubDate><description><![CDATA[1）

参考：https://zhuanlan.zhihu.com/p/82073864

SG-Net: Syntax-Guided Machine Reading Comprehension

https://arxiv.org/pdf/1908.05147.pdf

主要的思路是这样的：之前的话很多工作是用attention机制把所有的word算一遍，他们会觉得这样做没有考虑依存文法，他们尝...]]></description><category></category></item><item><title><![CDATA[dialog如何利用context知识]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/103300394</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/103300394</guid><author>yagreenhand</author><pubDate>Thu, 28 Nov 2019 21:50:01 +0800</pubDate><description><![CDATA[对话里面，有些模型是用了完整的context，有些模型是只用了上一句（我没有做完整统计），
1.到底哪一种更好呢？
2.如果采用第一种，能不能用好呢？

Do Neural Dialog Systems Use the Conversation History Effectively? An Empirical Study
Sharp Nearby, Fuzzy Far Away How Neur...]]></description><category></category></item><item><title><![CDATA[复现]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/103295752</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/103295752</guid><author>yagreenhand</author><pubDate>Thu, 28 Nov 2019 16:22:02 +0800</pubDate><description><![CDATA[1129开始

]]></description><category></category></item><item><title><![CDATA[latent4SLU]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/103234160</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/103234160</guid><author>yagreenhand</author><pubDate>Mon, 25 Nov 2019 20:23:36 +0800</pubDate><description><![CDATA[In this study, we aim to learn the meaningful distributed semantic representation, rather than focusing on building the NLU system itself. Once we obtained a reliable and reasonable semantic represen...]]></description><category></category></item><item><title><![CDATA[context建模学习(time-decay)]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/103216003</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/103216003</guid><author>yagreenhand</author><pubDate>Sat, 23 Nov 2019 20:24:29 +0800</pubDate><description><![CDATA[Because the framework can be trained in an end-to-end manner, all parameters (wi, a, b, e, f, D0, n) can be automatically learned to construct a flexible time-decay function.
具体阐释universal-&gt;flexibl...]]></description><category></category></item><item><title><![CDATA[slot-carryover学习]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/103211880</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/103211880</guid><author>yagreenhand</author><pubDate>Sat, 23 Nov 2019 11:26:39 +0800</pubDate><description><![CDATA[算是NLU和DST的结合。因为NLU只注意ut的slot和intent，而DST关注value的追踪（一般paper中slot是已经给定）。
1）Contextual Slot Carryover for Disparate Schemas
2）Improving Long Distance Slot Carryover in Spoken Dialogue Systems
要解决的问题：cont...]]></description><category></category></item><item><title><![CDATA[Memory4Dialog系列学习]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/103200045</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/103200045</guid><author>yagreenhand</author><pubDate>Fri, 22 Nov 2019 20:06:46 +0800</pubDate><description><![CDATA[记得上半年看GLMP，当时还跟别人讨论说memory都是骗人的，但是现在心态变正常了，能抓老鼠的都是好猫。下面对于这一系列的文章进行讲解。


1）LEARNING END-TO-END GOAL-ORIENTED DIALOG
2）Mem2Seq: Effectively Incorporating Knowledge Bases into End-to-End Task-Oriented Di...]]></description><category></category></item><item><title><![CDATA[domain-slot-value文章学习]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/103152320</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/103152320</guid><author>yagreenhand</author><pubDate>Tue, 19 Nov 2019 21:56:44 +0800</pubDate><description><![CDATA[主要针对mutlimoz2.0数据集进行学习
trade
submt
comer
ds-dst
som-dst
流行方法：建模context成sequential形式，单个domain-slot同时作为输入-----&gt;生成或者检索，
生成式：trade、comer、som-dst
检索式：submt、
生成检索结合：ds-dst
trade:

slot-gate是指，是否需要b的gener...]]></description><category></category></item><item><title><![CDATA[SF&ID文章]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/103141973</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/103141973</guid><author>yagreenhand</author><pubDate>Tue, 19 Nov 2019 16:08:05 +0800</pubDate><description><![CDATA[CM-Net：
过去的方法：
1.slotfilling和intentdetect分开的，rnn或者cnn建模语句，之后用分类算法
2.slotfilling和intentdetect是有关系的，

Goo2018intent影响slotfilling（intent产生之后在slot生成的LSTM中加元素）
zhang2018提出capsule，word-&gt;slot-&gt;intent，...]]></description><category></category></item><item><title><![CDATA[运行python程序输入到特定文件]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/102861514</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/102861514</guid><author>yagreenhand</author><pubDate>Fri, 01 Nov 2019 17:16:52 +0800</pubDate><description><![CDATA[logging教程

学习https://www.jianshu.com/p/feb86c06c4f4
]]></description><category></category></item><item><title><![CDATA[torch常用损失函数]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/102582261</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/102582261</guid><author>yagreenhand</author><pubDate>Wed, 16 Oct 2019 11:21:28 +0800</pubDate><description><![CDATA[https://www.cnblogs.com/marsggbo/p/10401215.html

总是记不住.....也是绝了.....
]]></description><category></category></item><item><title><![CDATA[qa论文精读目录——大佬推荐]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/102530359</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/102530359</guid><author>yagreenhand</author><pubDate>Sun, 13 Oct 2019 10:41:14 +0800</pubDate><description><![CDATA[1) Qingyu Zhou, Nan Yang, Furu Wei, Chuanqi Tan,Hangbo Bao, and Ming Zhou. 2017.  Neural ques-tion generation from text:  A preliminary study.InNational CCF Conference on Natural L...]]></description><category></category></item><item><title><![CDATA[问答系统概要]]></title><link>https://blog.youkuaiyun.com/yagreenhand/article/details/102527291</link><guid>https://blog.youkuaiyun.com/yagreenhand/article/details/102527291</guid><author>yagreenhand</author><pubDate>Sat, 12 Oct 2019 21:46:44 +0800</pubDate><description><![CDATA[Question Answering：

发现几位专家的看法，QA可以做的范围广，但是dialog的难度确实大，目前没什么大进展。对于VQA有很多方向可以发展。

Inferential Machine Comprehension: Answering Questions by Recursively Deducing the Evidence Chain from Text
Textbook Q...]]></description><category></category></item></channel></rss>